Who’s a Good Candidate?
How Party Gatekeepers Evaluate Potential Nominees∗
David Broockman
University of California, Berkeley
Nicholas Carnes
(corresponding author, nwc8@duke.edu)
Duke University
Melody Crowder-Meyer
Sewanee: The University of the South
Christopher Skovron
University of Michigan
August 20, 2014
WORKING PAPER
Abstract
Party leaders play a central but dimly understood role in shaping the pool of candidates voters
can choose from. Although many political observers blame these gatekeepers for exacerbating
polarization by recruiting extremists and for discouraging women and blue-collar workers from
running for office, we know little about how party leaders actually evaluate potential candidates
for their party’s nomination. In this paper, we shed new light on what these gatekeepers look for
in candidates with an original survey of roughly 6,000 leaders of county-level political parties.
We find that party leaders prefer candidates with the valence characteristics voters find
appealing—e.g., honesty, intelligence, dedication—suggesting that they play an important role in
recruiting the kinds of leaders the electorate wants. However, party leaders also use their
influence to discourage moderates from seeking office: they strongly prefer candidates at least as
ideologically polarized as their median party member. Republican party leaders show this
preference especially. Last, party leaders exhibit some biases against blue-collar workers but
show little bias against women. These findings have important implications for research on
candidate recruitment, party polarization, and the underrepresentation of women and the working
class.
The authors are grateful for financial support from the Russell Sage Foundation; for administrative support from
Belinda Keith; and for research assistance from Chris Kachadoorian, Jacob Zionce, Helen Chananie, Peter Chapin,
Allison Eisen, Taylor Festa, Alex Ghaffari, Anna Koelsch, Mike Margolis, Robin Millican, Jeremy Moore. David
Broockman and Chris Skovron also acknowledge support from the NSF Graduate Research Fellowship Program.
The authors are listed in alphabetical order. ∗
Politicians’ personal beliefs and backgrounds—such as their race, gender, class, ideology,
and experience—powerfully influence public policy and political representation in the United
States (Broockman 2013; Carnes 2013; Mansbridge 1999; Washington 2008). Yet the candidates
who voters choose from rarely reflect the full spectrum of demographics and political views
present in the public. Relative to voters and ordinary citizens, candidates for public office are
more likely to be white, male, wealthy, educated, and ideologically polarized.
Why is the pool of candidates for office in the United States so different from the people
who elect them? Scholars have theorized about both supply-side and demand-side influences on
the composition of the candidate pool. Some explanations emphasize patterns in the supply of
candidates interested in running for office. Others emphasize the demand for certain kinds of
candidates among voters, party elites, and other gatekeepers. Existing research has been
especially useful in illuminating supply-side factors like why different kinds of people are more
or less interested in running for office (Lawless 2011; Thomsen 2014). In this paper, we
complement this research with new evidence concerning a dimly understood feature of the
demand side of the candidate entry process: what kinds of people party leaders encourage to run
and support for their party’s nomination.
Leaders of political parties have a great deal of influence over the kinds of candidates
who appear on our ballots. They convince potential candidates to run (Broockman 2014;
Crowder-Meyer 2010b; Fox and Lawless 2010) and support their campaigns with resources and
endorsements (Bawn et al. 2012; Masket 2011; Masket and McGhee 2013). These leaders also
have considerable negative power, keeping candidates they dislike out of races (Niven 1998;
Sanbonmatsu 2006). Their decisions are ultimately crucial in shaping the field of choices
available to voters on election day.
1
However, party leaders can be difficult to study with readily available data because many
of their most influential choices are hidden from public view. When a party leader decides not to
encourage an ideological moderate to run for office because her views aren’t close enough to the
party’s positions, their decision does not leave a trace in standard political science datasets.
When a party primary is full of ideologically extreme candidates, it is thus difficult to know
whether it is because only extremists were interested in running in the first place or because
gatekeepers recruited extremists and discouraged moderates.
To shed light on how party leaders influence the candidate pool, we conducted a largescale, multi-mode survey of over 6,000 chairs of county-level branches of the Republican and
Democratic parties. Our survey used a variety of strategies to measure what party leaders look
for when they decide who to encourage to run for public office. We also administered a similar
survey to a nationally representative sample of ordinary Americans, which allowed us to
compare what party leaders look for in a candidate to what voters want in a politician.
Our findings—consistent across a variety of measurement strategies—help address
several important questions about the kinds of people who run for and win elected office in the
United States. First, our results show that party leaders strongly prefer candidates with the
valence characteristics that appeal to voters, such as honesty, intelligence, and work ethic. Party
leaders seem to play an important role in internalizing the collective demands of the electorate:
party leaders know that their side can only win elections if they nominate candidates with the
general characteristics that voters want, and they seem to work hard to recruit and support those
kinds of candidates.
However, our results also suggest that party elites play a crucial role in reinforcing
political polarization. Although parties appear to face electoral costs for nominating extreme
2
candidates (Hall forthcoming), the leaders of both parties nevertheless show a strong preference
for candidates who are more ideologically polarized. Democratic party leaders dislike candidates
who are more conservative than their median party member but do not mind candidates who are
more liberal than the median; Republicans party leaders dislike candidates who are more liberal
than their median party member but do not mind candidates who are more conservative. These
patterns are particularly pronounced among Republican elites, who place more than twice as
much weight on ideological fit as Democrats—and who appear to prioritize ideological fit above
any other characteristics of candidates. These findings provide some of the strongest evidence to
date that party leaders—especially in the Republican party—play an active role in encouraging
ideologically extreme candidates to run for public office.
Our survey also sheds new light on how party leaders influence the gender and social
class makeup of Americans political institutions. Women and working-class Americans are
sharply underrepresented in public offices at all levels, but it is unclear whether the cause has
more to do with demand from party leaders or a limited supply of interested candidates. In our
survey, party leaders exhibited some biases against working-class candidates but showed few
biases based on gender (or other factors associated with class and gender, like having a flexible
work schedule). These findings are consistent with a growing literature suggesting that class and
gender disparities in the candidate pool are not simply the result of differences in political
recruitment, but may also reflect forces that occur long before the candidate entry process, such
as socialization and political involvement early in life (Crowder-Meyer and Lauderdale 2014;
Lawless and Fox 2005; 2010; Shames 2014; Preece, Stoddard, and Fisher 2014).
Scholars have always understood that party leaders play a significant role in shaping the
candidate pool, but exactly how they do so has been a murky subject in political science. Our
3
findings illustrate how party leaders use their unique influence in the candidate emergence
process to both meet and frustrate the general electorate’s demands. On the one hand, party
leaders focus on recruiting the talented candidates voters want to serve in government. But they
also screen candidates along ideological and other lines, leaving the electorate with a
circumscribed set of choices that does not perfectly reflect what voters really want.
How Party Leaders Shape the Candidate Pool and Why It Matters
In the last few years, a growing body of research has recognized that politicians’ personal
beliefs and backgrounds play an important role in shaping the choices they make in office.
Characteristics like gender (Chattopadhyay and Duflo 2004), race (Broockman 2013), class
(Carnes 2012; 2013), religion (Burden 2009), and even having children (Washington 2008) all
seem to affect the kinds of policies leaders support and how they represent their constituents.
These findings have naturally led scholars to ask follow-up questions about why people
with certain characteristics are more or less likely to end up in office in the first place.
Candidates tend to be much more ideologically polarized than their constituents, yet ideological
moderates rarely run for office (Bafumi and Herron 2010). Politicians from the working class
tend to more reliably support many policies voters want, yet candidates from the working class
are exceedingly rare (Carnes forthcoming). Why do some kinds of people so seldom run for
public office?
Political scientists have learned an impressive amount about one set of explanations, the
factors that affect the supply of candidates interested in running for office. We know, for
instance, that gender differences in interest in candidacy help explain why so few women hold
4
office (e.g., Preece, Stoddard, and Fisher 2014) and that candidates who are more ideologically
in-step with their parties are more likely to want to run for office (Thomsen 2014).
We know less about another set of factors: differences in the demand for candidates
among party leaders and other political gatekeepers. However, there is plenty of cause to take
those explanations seriously. A growing body of research suggests that party elites play an
important role in shaping who the parties nominate for elected offices (e.g., Bawn et al. 2012).
First, party leaders play an important role in directing resources towards primary candidates they
support. Candidates who win a party’s nomination are the face of the party to voters, so party
leaders dedicate substantial effort to supporting the candidates who they think portray the party
in the best light and are the most likely to win (Bendavid 2007). They also provide endorsements
in primaries that many voters appear to see as influential cues (Masket and McGhee 2013).
Second, party leaders also appear to play an important role in recruiting candidates for
office who otherwise would not seek their party’s nomination at all (Crowder-Meyer 2010b;
Lawless 2012). In this form of recruitment, party leaders reach out to potential candidates and
encourage them to seek office, much like parties reach out to citizens to encourage them to vote.
As Broockman (2014) illustrates, this kind of recruitment is commonplace: at all levels of
government around the world, most politicians report that they were recruited into public service,
and most party leaders report that they routinely recruit candidates for public office. In our own
survey (described below), most party leaders indicated that candidates for office in their area
almost never simply decide to run for office on their own. True self-starters seem to be a
minority: party leaders and other gatekeepers usually have to seek out qualified candidates and
actively encourage them to seek office.
5
But what exactly are party leaders and other gatekeepers looking for in a candidate?
Scholars have written volumes about how voters choose between a few options on election day,
but we know very little about how gatekeepers sift through the thousands of eligible candidates
they could potentially support in the months and years beforehand.
Knowing what party leaders look for could have important implications for at least four
major areas of research on political candidates: ideology, human capital, class, and gender.
Ideology. Political observers have increasingly come to credit (and blame) party leaders
for shaping the ideological composition of the candidate pool. Authoritative scholars of
polarization often point fingers at the “political class”—and especially party leaders—for
supporting candidates with extreme views (Fiorina and Abrams 2009; Mann and Ornstein 2012).
However, it is by no means obvious what ideological criteria party leaders might be looking for
in candidates. On the one hand, party leaders have a strong incentive to nominate more moderate
candidates, since extreme nominees tend to lose elections (Canes-Wrone et al. 2002; Hall
forthcoming). Bendavid (2007) finds, for instance, that national Democratic elected officials tend
to favor nominating moderates who could win general elections. On the other hand, party leaders
are also likely to be ideologically polarized themselves (Crowder-Meyer 2010a), like other
political activists (Bafumi and Herron 2010). As such, they may prefer to nominate more
polarized candidates, even at the risk of losing a few votes. (This would be consistent with
research indicating that recruiters prefer to recruit candidates who are demographically similar to
themselves, e.g., Hunt and Pendley 1972; Prewitt 1970; Niven 1998).
Human capital. Studies of elections consistently find that voters prefer candidates with
talents, characteristics, and experiences associated with human capital (Buttice and Stone 2012;
Funk 1996; Hall and Bonneau 2006; Stokes 1963; Stone, Maisel, and Maestas 2004; Stone and
6
Simas 2010). Do party leaders similarly prioritize human capital when they recruit candidates,
either in the hopes of winning votes or in the hopes of fielding capable candidates?
Gender. In the last few years, scholars have paid special attention to party leaders’
preferences regarding candidate gender. There are signs that party leaders and other gatekeepers
may be at least partly responsible for the shortage of women in office: party elites are more
likely to ask men to run for office than women (Crowder-Meyer 2013; Lawless and Fox 2005;
Niven 1998). On the other hand, party leaders’ recruitment outcomes may reflect gender
differences in who is interested in seeking office in the first place (Lawless and Fox 2010).
Gender disparities in the candidate pool seem to have roots that begin before the candidate entry
process, such as in socialization and political involvement (Crowder-Meyer and Lauderdale
2014; Lawless and Fox 2013). For example, Preece, Stoddard, and Fisher (2014) find that
politically active women were half as likely to respond to a party’s invitation to learn how to run
for office as politically active men. If men and women showed equal levels of interest in office,
it is conceivable that party leaders might recruit equal numbers of men and women to run.
Class. Building on work on the gender composition of the candidate pool, scholars
interested in the shortage of candidates from the working class have also recently speculated that
party leaders and other gatekeepers may be biased against candidates from the working class,
that is, from manual labor and service industry jobs (e.g., Carnes 2013, ch. 6). There seem to be
many qualified working-class citizens, and in places where workers are well-integrated into party
organizations and labor unions, workers tend to hold office in larger numbers (Carnes 2014;
Sojourner 2013). Of course, there are many other factors that might explain why so few workers
run and win. Moreover, no study to date has ever investigated whether party leaders and other
gatekeepers are biased against workers.
7
Simply put, scholars still need to learn a great deal about how party leaders decide who to
encourage to run for public office. As a step in that direction, we have simply asked them.
The National Survey of Party Leaders
In November 2013, we fielded the National Survey of Party Leaders, a self-completed
survey of the chairs of the roughly 6,000 county-level (or equivalent)1 branches of the
Republican and Democratic parties.
We began by collecting their email and/or physical mailing addresses from publicly
available sources. (In 9 states, neither party posts contact information for county-level party
officials online: Georgia, Indiana, Iowa, Kentucky, Michigan, New Hampshire, New Mexico,
Oklahoma, and Wisconsin. We excluded those states from our analysis.) We then sent postcards
and pre-survey emails to each respondent (if we had both a mailing address and an email
address, we sent both), then followed up a week later with a full letter and/or email inviting the
chair to complete the survey. Of the 6,219 chairs we contacted, we received responses from
1,118 (18%), a response rate comparable to recent mail surveys of sitting politicians (e.g.,
Broockman and Skovron 2013), although somewhat lower than gatekeeper surveys conducted in
the past (e.g., Crowder-Meyer 2010).
To determine whether party leaders look for the same kinds of qualities that voters value,
in January 2014, we conducted a parallel survey of ordinary citizens that asked many of the same
kinds of questions. The survey was fielded by Survey Sampling International, which administers
surveys to pools of online respondents selected to resemble the demographic makeup of the
country as a whole. The survey was completed by 1,240 respondents, giving us a large enough
sample to make generalizations about the population as a whole.
1
Louisiana’s parties are organized by parish, Alaska’s are organized by borough, North Dakota’s are organized by
district, Connecticut’s are organized by city, and the Democratic party in Massachusetts is organized by sub-city
unit.
8
On both surveys, we included several items designed to measure the traits and qualities
that respondents look for in an ideal candidate. To ensure that we reached robust conclusions, we
used three different types of questions: multiple-choice questions, open-ended items, and a
survey experiment describing hypothetical candidates.
The multiple choice questions were designed to encourage respondents to directly
evaluate the relative importance of several traits that voters value. These questions began, “In
your opinion, how important is it that a nominee for elected office from your party . . .” and were
then broken into four categories, “. . . have the following qualifications?,” “. . . be from the
following professional backgrounds?” “. . . have the following personal qualities?”, and “. . .
have the following life circumstances?” Within each category, we inquired about three to five
specific traits. Respondents could answer, “Not important,” “Somewhat important,” or “Very
important.”
To check that our multiple-choice questions had not overlooked other traits that are
important to respondents, on our survey of party leaders, we also asked a simple open-ended
question, “In an ideal world, what personal qualities would you like all your party’s political
candidates to have? Please list as many as you would like.” We then coded respondents’ answers
into roughly three dozen categories.
Finally, to check that respondents’ stated preferences would persist when they were
forced to make choices between candidates, we also included a hypothetical candidate survey
experiment on both the survey of party leaders and the survey of the public. The experiment used
a conjoint design (Hainmueller, Hopkins, and Yamamoto 2014; Hainmueller, Hangartner, and
Yamamoto 2014): we provided each respondent with brief descriptions of two hypothetical
9
candidates, randomizing the traits of each candidate, and then asked each respondent, “Who
would you be more likely to encourage to run for office?”
At the beginning of the experiment, we first told respondents, “Suppose there is a primary
for an open seat in your county and the two individuals below are considering running. We’d like
you to consider the following two potential candidates for this office.” We then supplied each
candidate’s name (randomizing whether the respondent saw a male or female first name), age
(randomizing whether the respondent saw a younger or older candidate), occupation
(randomizing whether the respondent saw jobs in law and business, politics and education, or
working-class jobs), experience in the party (randomizing whether the respondent saw a more or
less dedicated party activist), life circumstances (randomizing among having free time, being a
veteran, having flexible work hours, being independently wealthy, or having two young
children), talents (randomizing among the traits of assertiveness, fundraising experience, work
ethic, physical attractiveness, public speaking ability, and notoriety in the community), and
political ideology (randomizing among being similar to the typical constituent, somewhat more
liberal, somewhat more conservative, much more liberal, and much more conservative). This
approach allowed us to measure the relative importance of a wide range of characteristics in a
setting where respondents—either voters or party leaders—had to make a choice between two
competing options.
Together, these analyses gave us three different ways to ask the ordinary citizens who
elect politicians and the party gatekeepers who recruit them the same fundamental question:
“What do you look for in a candidate?”
What Party Gatekeepers (And Voters) Look For In Candidates
10
Our findings were remarkably consistent across these three measures—and between our
samples of party leaders and voters. Both party leaders and voters strongly preferred candidates
with valence characteristics like honesty, intelligence, and dedication. Party leaders also
exhibited a strong preference for more ideologically polarized candidates, especially Republican
party leaders. The party leaders in our survey also exhibited some biases against blue-collar
workers but showed essentially no bias against women. What do party leaders (and voters) look
for in a candidate? Personal talents and ideological orthodoxy.
Multiple Choice Questions
In the survey, respondents were given the option to rate of several traits either “not
important,” “somewhat important,” or “very important. To simplify the presentation, Figure 1
simply averages their responses (treating not important as zero, somewhat important as a one,
and very important as two.) The figure groups the traits the way respondents were asked about
them, with traits related to candidates’ qualifications first (e.g., “has previously held office”),
their professional background second (e.g., “is a businessperson”), their personal qualities third
(e.g., “is assertive”), and their life circumstances fourth (e.g., “has a flexible work schedule”).
Viewed this way, several patterns are immediately apparent. First, citizens and party
leaders gave remarkably similar answers. The ordinary citizens who responded to our public
opinion survey (depicted with grey bars in the figure) rated most of the traits on our survey about
as important or unimportant as the party chairs who responded to our National Survey of Party
Leaders. There were a few notable exceptions (discussed in more detail in the next section). But
overall, party leaders appeared to define a good candidate in roughly the same way that ordinary
citizens did.
11
Figure 1: What Traits Do Party Leaders and Citizens Think Candidates Should Have?
0.48
Has previously held office
Is an active political activist / volunteer
Is an active community activist / volunteer
Has held a community leadership position
Has close ties with groups important to the party
0.78
1.021.13
1.05
0.96
1.08
0.99
Is a businessperson
Is an educator
Is a lawyer
0.64
0.45
0.55
Party Leaders
1.32
1.30
Citizens
0.85
0.91
0.86
1.391.53
Is assertive
Is a hard worker
Is personable
Is physically attractive
0.38
Has a flexible work schedule
Has a great deal of free time
Is independently wealthy
Able to raise money from friends and associates
0.36
0.42
1.87
1.84
1.83
1.65
0.64
0.810.92
1.00
0.0
Not important 0.5
1.0
Somewhat important 1.28
1.26
1.33
1.5
2.0
Very important What traits were important? The only traits that cleared the 1.5 mark—that is, that were
thought to be very important by a majority of respondents—were being assertive, being a hard
worker, and being personable. To voters and party leaders alike, a good candidate is not defined
by reputation, occupation, or campaign credibility per se; she’s defined by human capital traits
like being dedicated and friendly.
Next in importance were political experience variables. While the average rating for most
of these traits was close to “somewhat important,” the various kinds of experience as a political
or community activist that we asked about all received some attention from both the public and
the party leaders. Notably, experience as an elected official received low rankings, particularly
12
among the party leaders; rather, party leaders and voters seem to be looking for general
experience and familiarity with the community.
Social class markers were generally the least important characteristics in this analysis.
Respondents rated working in business, law, and education as low in importance; these traits
averaged slightly less than “somewhat important” in both samples. When asked to directly rank
the importance of work in three of the “pipeline professions” from which most actual politicians
are drawn, both party leaders and voters responded with a shrug.
Open-ended Responses
We also asked party leaders to describe their ideal candidates without imposing any
structure on their responses. Strikingly, our open-ended question elicited the same basic
responses as our multiple-choice items.
The print version of our party leader survey included an item that simply asked what
characteristics party leaders wished candidates would have in an ideal world. Respondents could
list as many as they wanted. Of the respondents who completed a paper version of the survey,
234 (84%) listed at least one characteristic. Figure 2 plots the numbers of respondents who listed
each of 37 traits or characteristics.
Like their multiple-choice answers, the characteristics party leaders mentioned in
response to this open-ended question largely focused on general human capital—personal
qualities and experiences—rather than specific occupational or life circumstances. Only 4%
mentioned a background in business, and no other occupation came up. Only 1% mentioned
gender. Only 2% mentioned being politically active. Only 5% mentioned formal educational
credentials. By far, the most common responses party leaders gave focused on the kinds of
13
human characteristics that predict success in life in general: honesty (41%), intelligence (37%),
friendliness (35%), work ethic (30%), and communication skills (24%).
Figure 2: Open-ended Response Counts
honest intelligent personable hardworking good at speaking confident good at fundraising conserva;ve listens involved in the community moderate enjoys campaigning experienced concerned about good public policy loyal to the party cons;tu;onal conserva;ve person of faith concerned about cons;tuents compassionate educated aKrac;ve liberal affluent / business background other / vague poli;cally ac;ve humble uncontroversial has a family funny has lots of free ;me long-­‐;me resident young female affiliated with unions op;mis;c not influenced by money patrio;c 97 87 82 70 55 34 30 29 25 23 22 21 21 18 17 16 16 15 14 14 12 12 12 7 7 6 6 6 5 4 3 2 2 2 1 1 1 0 20 40 60 80 100 120 14
The open-ended question also gave party leaders the opportunity to discuss ideological
considerations, which we did not include in our multiple choice items. Their open-ended answers
suggested that ideology figures very prominently in their choices. In total, 34% of party leaders
mentioned that an ideal candidate was someone who was liberal, conservative, moderate, or loyal
to the party’s positions.
Importantly, the responses to these open-ended questions were not symmetric by party.
Figure 3 plots the proportions of Republican and Democratic party leaders who mentioned each
of the five most common traits from Figure 2 as well as the proportion who mentioned ideology
(here we simply sum any mention of ideology, liberal or conservative). Strikingly, Republican
party leaders were twice as likely as Democrats to mention ideology in their open-ended
responses. Close to half of Republican party leaders discussed ideology as an important
characteristic of their ideal candidate, whereas only about 20% of Democrats thought of their
ideal candidates in ideological terms.
Figure 3: The Proportion of Republican and Democratic Party Leaders Who Mentioned
Each of the Six Popular Traits in an Open-ended Ideal Candidate Question
0.5 Democrat 0.45 Republican 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 Honest Intelligent Personable Hardworking Good Communicator Ideology 15
Together, the results of the open-ended and multiple-choice questions painted a
consistent picture: when party leaders evaluate potential candidates, they focus largely on
markers of talent (just as voters do) and on ideology. As one Republican party chair wrote in
response to our open-ended question: What would his party’s candidates look like in an ideal
world? “They would all be Ronald Reagan.”
Survey Experiment
Our hypothetical candidate survey experiment reached the same basic conclusions.
Experiments presenting hypothetical candidates are nothing new in studies of candidate
attributes, but our experiment was unique in two important respects. First, we used a conjoint
analysis (Hainmueller, Hopkins and Yamamoto 2014), which allowed us to randomize across a
variety of attributes that may influence candidate evaluations and recover estimates of the effect
of each. Second, we conducted the same experiment in our survey of county party leaders and in
our survey of the general public, which allowed us to directly compare how important each
characteristic was to both citizens and party leaders. More broadly, these experiments allowed us
to measure what our respondents look for in a candidate in a way that was less subject to social
desirability biases and other distortions that can sometimes occur in explicit survey questions.
We altered several aspects of the candidates: their age, gender, ideology, occupation,
previous experience, life circumstances, and ideology. The text of the manipulations are given in
the Appendix. Note that for the ideology manipulation, we manipulated the candidate’s
ideological position relative to “the typical voter in your county from your party,” not in general.
16
Figure 4 plots the average effect (the component-specific marginal effect, or ACME) of
each candidate characteristic on the probability that a party leader who would encourage the
candidate to run for public office. Figure 5 displays the same estimates from our survey of
ordinary citizens.
Our treatment categories randomized the candidates’ ages but kept them similar (43 or
47), and not surprisingly, we found that the difference had no effect on whether party leaders or a
member of the public would encourage the candidate to run for office. Far more surprising was
that gender had essentially no effect in our survey of ordinary citizens—and that party leaders
were actually slightly more likely to prefer female candidates to male candidates.
Figure 4. How Candidate Attributes Affect Support from Party Leaders
Age 43
Age 47
Male
Female
White collar 1 (Professional)
Blue collar
White collar 2
None
Active and well known in county party organization
Active and well known in group important to the party
Frequent campaign volunteer for last four election cycles
Frequent campaign volunteer in last election cycle
Has a great deal of free time
Has flexible work hours
Has two young children
Is independently wealthy
Military veteran
Assertive
Experienced fundraiser for local charities
Hard worker
Physically attractive
Talented public speaker
Well known in community
D. Average
R. Average
D. Conservative
R. Conservative
D. Liberal
R. Liberal
D. Very conservative
R. Very conservative
D. Very liberal
R. Very liberal
●
●
●
●
●
●
●
●
●
●
●
●
Attribute:
●
●
●
●
●
●
Age
●
Experience
●
Gender
●
Ideology relative to party
●
Life circumstances
●
Occupation
●
Talents
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
−0.6
−0.4
−0.2
0.0
0.2
0.4
Change: Pr(Encourage to Run)
17
Compared to a candidate who has had no involvement with the party in the last few
years, a candidate who has been a more frequent volunteer is more likely to be preferred by both
voters and party officials, as we might expect if we thought that Americans defined good
candidates as those who had held office before or those who are well-known to party officials.
In contrast to our open-ended and multiple-choice questions, our survey experiment also
provided some evidence that party leaders and voters prefer candidates from white-collar jobs to
those from blue-collar jobs. The estimated effects were relatively small, but statistically
significant.
Figure 5. How Candidate Attributes Affect Support from Voters
Age 43
Age 47
Male
Female
White collar 1 (Professional)
Blue collar
White collar 2
None
Active and well known in county party organization
Active and well known in group important to the party
Frequent campaign volunteer for last four election cycles
Frequent campaign volunteer in last election cycle
Has a great deal of free time
Has flexible work hours
Has two young children
Is independently wealthy
Military veteran
Assertive
Experienced fundraiser for local charities
Hard worker
Physically attractive
Talented public speaker
Well known in community
D. Average
R. Average
D. Conservative
R. Conservative
D. Liberal
R. Liberal
D. Very conservative
R. Very conservative
D. Very liberal
R. Very liberal
●
●
●
●
●
●
●
●
●
●
●
●
Attribute:
●
●
●
Age
●
●
Experience
●
Gender
●
Ideology relative to party
●
Life circumstances
●
Occupation
●
Talents
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
−0.6
−0.4
−0.2
0.0
0.2
0.4
Change: Pr(Encourage to Run)
Consistent with our other survey items, voters and party leaders said they were more
likely to vote for candidates who were hardworking. And just as we found in other items, by far
18
the biggest effects evident in Figures 4 and 5 were those pertaining to ideology. Republican
voters and party leaders were very unlikely to support a potential candidate who was liberal.
Democratic voters and party leaders were unlikely to support a potential candidate who was
conservative.
Importantly, ideology had a substantially larger impact on the candidate preferences of
party leaders than on the candidate preferences of voters. A candidate who was very liberal
relative to her party was 20% less likely to be preferred by a Republican in the general public,
but 40% less likely to be preferred by a Republican party leader.
Any way we asked what they looked for in a candidate, political elites and ordinary
citizens gave us essentially the same answers. Like voters, party leaders prefer candidates who
are talented and capable. And like voters, party leaders prefer candidates who share the party’s
ideology—but they care more.
Implications for Political Representation
These findings have three important implications for political representation. First, party
leaders appear to play an important role in helping select candidates for office who have high
levels of human capital. Candidates for political office are significantly more likely to have
leadership experience, be highly educated, and have other traits that are appealing to voters and
likely to translate into superior job performance. Party leaders appear to anticipate voters’
demands for these traits and focus on recruiting candidates with them.
At the same time, party leaders use their unique role in the candidate emergence process
to exert strong ideological screening. It is well-established that politicians today are more highly
polarized than ever, with most supporting a consistent set of policies in line with their party’s
19
platform. Our results suggest that party leaders play an important role in generating this
polarization, as leaders in both parties strongly prefer candidates who are at least as extreme as
their party’s median voters and show strong distaste for candidates closer to the general median
voter.
Our results also point to party leaders as contributing to asymmetric polarization.
Research has shown that Republican elites, especially members of Congress, have polarized
more sharply than their counterparts on the political left (e.g., Hacker and Pierson 2005). Our
findings help explain this pattern, as they suggest that Republican party chairs and, to a lesser
extent, Republican primary voters, are weighing ideology more in the candidate selection stage,
leading to a pool of Republican candidates that is more ideologically staunch than the
Democratic pool. The preference for very conservative candidates seems stronger among party
leaders than among voters, although we cannot tell from our data whether one group has led the
other to adopt this preference.
The responses to the open-ended results demonstrate this trend. Republicans are much
more likely to, unprompted, mention ideology as an important factor for candidates. Our
evidence suggests that not only do Republicans care more about ideology, it is also readily
accessible when they think of candidate recruitment. It seems likely, then, that Republicans are
much more active in recruiting ideologically polarized candidates than Democrats are.
The results of our survey experiment put this trend in stark relief. In general, ideology
matters much more than other attributes in shaping party chairs’ decisions of which candidates to
support. And, this focus on ideological recruitment contributes to polarization, as neither party’s
chairs are likely to support a candidate more moderate than the party median. However, there is
an ideological asymmetry which suggests that the effects of recruitment on polarization are not
20
equal across the two parties. Democratic chairs are most inclined to support candidates who are
middle-of-the-road or slightly left with respect to the party, while Republicans prefer candidates
who espouse an ideology matching or more conservative than their party. In fact, while
Democratic chairs are less likely to support very liberal candidates than those nearer to their
party average, Republican chairs seem to give very conservative primary candidates the same
boost that Democrats give to moderates. While we do not have enough evidence to contend that
candidate recruitment practices cause elite polarization, this evidence suggests that the actions of
the parties are possible culprits.
Finally, our results square with other research suggesting that party leaders and other
gatekeepers are not entirely to blame for gender and social class gaps in officeholding in the
United States. We find no evidence that party leaders prefer male candidates, nor do we find any
evidence that they have a preference for particular white-collar jobs. We do find that party
leaders and voters alike are less likely to recommend that a candidate from the working class run
for public office (and some of the personality traits that elites and citizens prefer may be easier
for men and white-collar professionals to obtain). But the size of the penalty against workers in
our survey experiment is modest compared to the magnitude of the shortage of working-class
citizens in public office. Simply put, something other than the preferences of party leaders seems
to be at work in the persistent underrepresentation of women and workers in American political
institutions. (Of course, party leaders could still help to increase the economic and gender
diversity of the candidate pool by actively recruiting more women and workers.)
More broadly, our analysis underscores the need for more research not just on elections
and governing, but on the candidate emergence and political gatekeeping processes that occur
beforehand. Voters and party leaders generally have similar ideas about what makes a good
21
candidate, but our analysis also uncovered important differences. Are there others? How do party
leaders actually do the hard work of identifying and recruiting candidates? Candidate recruitment
is a widespread practice in US politics, and political scientists have only begun to understand its
implications for American democracy.
Bibliography
Bafumi, Joseph and Michael Herron. 2010. “Leapfrog Representation and Extremism: A Study
of American Voters and Their Members of Congress.” American Political Science Review
104: 519-542.
Bawn, Kathleen, Martin Cohen, David Karol, Seth Masket, Hans Noel, and John Zaller. “A
Theory of Political Parties: Groups, Policy Demands, and Nominations in American
Politics.” Perspectives on Politics 10(3): 571–597.
Bendavid, Naftali. 2007. The Thumpin’. New York: Doubleday.
Broockman, David E. 2013. “Black Politicians Are More Intrinsically Motivated To Advance
Blacks’ Interests: A Field Experiment Manipulating Political Incentives.” American Journal
of Political Science 57(3): 1–16.
Broockman, David E. 2014. “Mobilizing Candidates: A Field Experiment and a Review.”
Forthcoming, Journal of Experimental Political Science.
Broockman, David E., and Christopher Skovron. 2013. “What Politicians Believe About Their
Constituents: Asymmetric Misperceptions and Prospects for Constituency Control.” Working
Paper.
Burden, Barry C. 2009. The Personal Roots of Representation. Princeton: Princeton University
Press.
Buttice, Matthew K., and Walter J. Stone. 2012. “Candidates Matter: Policy and Quality
Differences in Congressional Elections.” The Journal of Politics 74(3): 870–887.
Canes-Wrone, Brandice, David W. Brady, and John F. Cogan. 2002. “Out of Step, Out of Office:
Electoral Accountability and House Members’ Voting.” American Political Science Review
96(1): 127–140.
Carnes, Nicholas. 2012. “Does the Numerical Underrepresentation of the Working Class in
Congress Matter?” Legislative Studies Quarterly 37(1): 5–34.
Carnes, Nicholas. 2013. White-Collar Government: The Hidden Role of Class in Economic
Policy Making. Chicago: University of Chicago Press.
22
Carnes, Nicholas. 2014. “Why Are There So Few Working-Class People in Political Office?
Evidence From State Legislatures.” Working Paper.
Carnes, Nicholas. Forthcoming. “Does the Descriptive Representation of the Working Class
'Crowd Out' Women and Minorities (And Vice Versa)? Evidence from the Local Elections in
America Project.” Politics, Groups, and Identities.
Chattopadhyay, Raghabendra, and Esther Duflo. 2004. “Women as Policy Makers: Evidence
from a Randomized Policy Experiment in India.” Econometrica 72(5): 1409–1443.
Crowder-Meyer, Melody A. 2010b. "The Party's Still Going: County Party Strength, Activity,
and Influence." In The State of the Parties: The Changing Role of Contemporary American
Parties, ed. J. Green and D. Coffey. Lanham, MD: Rowman and Littlefield.
Crowder-Meyer, Melody Ara. 2010a. “Local Parties, Local Candidates, and Women’s
Representation: How County Parties Affect Who Runs For and Wins Political Office.”
Dissertation. Princeton University, Princeton, NJ.
Crowder-Meyer, Melody, and Benjamin E. Lauderdale. 2014. “A Partisan Gap in the Supply of
Female Potential Candidates in the United States.” Research and Politics 1(1): 1–7.
Crowder-Meyer, Melody. 2013. "Gendered Recruitment without Trying: How Local Party
Recruiters Affect Women's Representation." Politics & Gender 9 (4).
Fiorina, Morris P. and Samuel J. Abrams. 2009. Disconnect: The Breakdown of Representation
in American Politics. Norman: University of Oklahoma Press.
Funk, Carolyn L. 1999. “Bringing the Candidate into Models of Candidate Evaluation.” Journal
of Politics 61(3): 700–20.
Hacker, Jacob S., and Paul Pierson. 2005. Off Center: The Republican Revolution and the
Erosion of American Democracy. New Haven: Yale University Press.
Hainmueller, Jens, Dan Hopkins, and Teppei Yamamoto. 2014. “Causal Inference in Conjoint
Analysis: Understanding Multi-Dimensional Choices via Stated Preference Experiments.”
Political Analysis 22(1): 1–30.
Hall, Andrew B. Forthcoming. “What Happens When Extremists Win Primaries?” American
Political Science Review.
Hall, Melinda Gann, and Chris W. Bonneau. 2006. “Does Quality Matter? Challengers in State
Supreme Court Elections.” American Journal of Political Science 50(1): 20–33.
Hunt, Jr., A. Lee and Robert E. Pendley. 1972. Midwest Journal of Political Science 16(3): 411438.
Lawless, Jennifer L. 2011. Becoming a candidate: political ambition and the decision to run for
office. Cambridge: Cambridge University Press.
23
Lawless, Jennifer L. and Richard L. Fox. 2013. “Girls Just Wanna Not Run: The Gender Gap in
Young Americans’ Political Ambition.” Washington, DC: Women & Politics Institute.
Lawless, Jennifer L., and Richard L. Fox. 2005. It Takes a Candidate: Why Women Don’t Run
for Office. New York: Cambridge University Press.
Lawless, Jennifer L., and Richard L. Fox. 2010. It Still Takes a Candidate: Why Women Don’t
Run for Office. New York: Cambridge University Press.
Mann, Thomas E., and Norman J. Ornstein. 2012. It’s Even Worse Than It Looks: How the
American Constitutional System Collided with the New Politics of Extremism. New York:
Basic Books.
Mansbridge, Jane. 1999. “Should Blacks Represent Blacks and Women Represent Women? A
Contingent ‘Yes.’” The Journal of Politics 61(3): 628–657.
Masket, Seth. 2011. “The Circus That Wasn’t: The Republican Party’s Quest for Order in
California's 2003 Gubernatorial Recall Election." State Politics and Policy Quarterly 11(2):
123-147.
Niven, David. 1998. “Party elites and women candidates: The shape of bias.” Women & Politics,
19(2), 57-80.
Niven, David. 1998. “Party Elites and Women Candidates: The Shape of Bias.” Women and
Politics 19: 57-80.
Prewitt, Kenneth. 1970. The Recruitment of Political Leaders: A study of citizen-politicians.
Indianapolis: Bobbs-Merrill.
Sanbonmatsu, Kira. 2006. Where Women Run: Gender and Party in the American States. Ann
Arbor: University of Michigan Press.
Shames, Shauna. 2014. “Roots of Political Ambition: Gender, Race, & the Decision (Not) to
Seek Office.” Dissertation. Harvard University.
Sojourner, Aaron. 2013. “Do Unions Promote Electoral Office-Holding? Evidence from
Correlates of State Legislatures’ Occupational Shares.” Industrial and Labor Relations
Review 66(2): 467-486.
Stone, Walter J., L. Sandy Maisel, and Cherie D. Maestas. 2004. “Quality Counts: Extending the
Strategic Politician Model of Incumbent Deterrence.” American Journal of Political Science
48(3): 479–495.
Washington, Ebonya. 2008. “Female Socialization: How Daughters Affect Their Legislator
Fathers’ Voting on Women’s Issues.” American Economic Review 98 (1): 311–322.
24
Appendix
Table A1. Attributes of candidates in conjoint experiment
Each of the two candidate profiles presented to the respondents was randomly assigned an attribute from
each box.
Attribute
Name (gender)
Age
Occupation
Experience in party
Life circumstances
Talents
Positions and ideology
Values
Male names: Donald, Laurence, Nathan,
Nicholas, Samuel, Alexander, Andrew,
Christopher, Charles, Daniel. Female
names: Donna, Lauren, Natalie, Nicole,
Samantha, Alexandra, Andrea, Charlotte,
Christina, Danielle. (No pair of candidates
had the same name.)
43, 47
Attorney, business executive, investor,
factory worker, lawyer, nurse, receptionist,
restaurant server, small business owner,
social worker, teacher
Active and well known in county party
organization, active and well known in
group important to the party, frequent
campaign volunteer for the last four
election cycles, frequent campaign
volunteer in last election cycle, none
Has a great deal of free time, has two
young children, has flexible work hours, is
independently wealthy, military veteran
Assertive, experienced fundraiser for local
charities, hard worker, physically
attractive, talented public speaker, well
known in community
Much more conservative than the typical
voter from your party in your county,
somewhat more conservative than the
typical voter from your party in your
county, similar views to the typical voter
from your party in your county, somewhat
more liberal than the typical voter from
your party in your county, much more
liberal than the typical voter from your
party in your county
25
These attributes were fully randomized. Other than name, the candidates were allowed to share attributes
across the two profiles, but each candidate had only one attribute from each box on the right side of Table
A1.
26